We have generated conditional glucocorticoid receptor (GR) knockout mice, which have been crossed to lck-Cre animals to produce mice that lack GR expression in thymocytes and T cells (GRlci-Cre). These mice have a modest reduction (about 40%) in the number of double positive (DP) and single positive (SP) thymocytes. Introduction of a transgenic T cell receptor (TCR) that has differing affinities for different MHC-encoded class II molecules revealed that the higher the avidity for self, the greater the reduction in thymocyte number in the GR KO mice, indicating that the thymocytes are undergoing increased negative selection in the absence of GR signaling. If the TCR repertoire, the range of receptors expressed after selection, is indeed affected by glucocorticoids, there should be changes in the specificity of immune responses. We have found that mature T cells from the conditional GR-null mice proliferate normally to mitogens or to TCR cross-linking, but poorly to alloantigen. Moreover, immunization with peptide antigens results in a poor T cell recall response. Strikingly, if the TCR repertoire is fixed by introducing alpha/beta TCR transgenes, the proliferative response of GRlci-Cre T cells to antigen in normal, in vitro and in vivo. Deep sequencing of TCRbeta CD3 regions, which have the largest contribution to TCR specificity, found that there was a significant difference between, but not within, groups of wild type and GRlci-Cre naive T cells. These results demonstrate that exposure to glucocorticoids in the thymus is a critical event in shaping the T cell repertoire, and thus the ability to respond to foreign pathogens. We have now made mice in which the enzyme responsible for corticosterone production, CYP11B1, encoding steroid 11beta-hydroxylase, is floxed. We have crossed these mice onto Cre-expressing animals to knockout glucocorticoid production in epithelial cells, allowing us to test the hypothesis that local glucocorticoid production is important for thymocyte development. Because glucocorticoids are also immunosuppressive, we have examined the immune response in several infectious models. In one such model, T. gondii, we found that mice lacking GR expression in T cells undergo a cytokine storm and die within days of infection (in contrast, wild type animals mount a functional immune response and survive). These data demonstrate that the response of peripheral T cells to endogenous glucocorticoids is non-redundant in controlling the limits of an immune response to prevent harmful secondary effects. Thus, glucocorticoids have opposing roles in the adaptive immune system, promoting the selection of a relative high affinity TCR repertoire required for immunological fitness, and preventing excessive effector cell function that can be deleterious to the host. Finally, we have knocked out the GR in DC. We find that in sepsis models, CD8+ DC are deleted (apoptotic death) by stress levels of glucocorticoids, and when this is prevented the mice become very ill due to cytokine storm (IL-12, IFNgamma, TNF). Furthermore, such mice are resistant to LPS tolerance, demonstrating that it is the loss of CD8+ DC that renders mice tolerant of subsequent challenge with LPS. Thus, a contribution of DC to sepsis is prevents by endogenous glucocorticoids, which is consistent with the fact that increased morbidity and mortality from sepsis is associated with adrenal insufficiency.